CN112749661A - Traffic accident responsibility judging model based on block chain and IVggNet - Google Patents
Traffic accident responsibility judging model based on block chain and IVggNet Download PDFInfo
- Publication number
- CN112749661A CN112749661A CN202110046331.9A CN202110046331A CN112749661A CN 112749661 A CN112749661 A CN 112749661A CN 202110046331 A CN202110046331 A CN 202110046331A CN 112749661 A CN112749661 A CN 112749661A
- Authority
- CN
- China
- Prior art keywords
- data
- ivggnet
- model
- accident
- block chain
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 206010039203 Road traffic accident Diseases 0.000 title claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000005516 engineering process Methods 0.000 claims abstract description 9
- 238000012937 correction Methods 0.000 claims abstract description 8
- 238000012545 processing Methods 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims description 15
- 238000011176 pooling Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 14
- 238000012544 monitoring process Methods 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 7
- 230000006855 networking Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 230000001133 acceleration Effects 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 3
- 230000036544 posture Effects 0.000 claims description 3
- 230000004044 response Effects 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims 1
- 238000013135 deep learning Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/27—Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Multimedia (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Strategic Management (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Technology Law (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Fuzzy Systems (AREA)
- Probability & Statistics with Applications (AREA)
- Traffic Control Systems (AREA)
Abstract
A traffic accident responsibility judging model based on a block chain and IVggNet. Step 1, uploading data of the traffic accident to a block chain node; step 2, classifying and integrating the uploaded accident data by using a block chain technology and correspondingly storing the uploaded accident data; step 3, taking data from the block chain by using the proposed IVggNet model; step 4, comparing the discrimination result in the step 3 with the responsibility discrimination result of the traffic police department; step 5, uploading the data of the accident and the judgment result of the traffic police department to a block chain model error correction training set, updating the IVggNet model when the sample data in the error correction training set reaches a set threshold value, and finally uploading the updated IVggNet model to a block chain to realize the sharing of the data and the model; and 6, processing the traffic accident according to the result of the judgment. Under the technical support of block chains and the like, the intelligent responsibility judgment of the traffic accident is realized through the proposed IVggNet network model.
Description
Technical Field
The invention relates to the field of block chains and road traffic, in particular to a traffic accident responsibility judgment model based on the block chains and IVggNet.
Background
With the development of urbanization construction and the development of national economy and technology level in China, the automobile holding capacity of residents in China increases year by year, and traffic accidents can inevitably happen under the condition that the automobile holding capacity is more and more. In 2018, 46161 people die from 166906 traffic accidents of automobiles, 169046 people are injured, and 118671.6 ten thousand yuan of direct property loss is caused. In recent years, road networks have been basically formed, and traffic monitoring systems which are important components of road traffic are more continuously optimized and perfected, so that main road sections are basically and completely covered, and powerful data guarantee is provided for the responsibility judgment of traffic accidents. The responsibility of the traffic accidents is mainly manual responsibility judgment at present, and a large number of traffic accidents bring huge workload, waste huge manpower and material resources and are not beneficial to the development of the society.
On one hand, in recent years, deep learning is developed rapidly, and various network models such as a convolutional neural network, an confrontation generation network and variants thereof are proposed by scientific researchers in large quantity and are widely applied; on the other hand, the performance of the deep learning model is greatly restricted by the number of training data, and how to acquire massive sample data and effectively utilize the data becomes a problem to be solved urgently. On the other hand, videos or pictures shot by the road monitoring equipment may be blurred in rainy and foggy weather, so that the algorithm model for determining responsibility cannot well extract features. Therefore, how to effectively store and utilize mass traffic accident data nationwide and even worldwide, and simultaneously, a proper local feature extraction algorithm is adopted, and an intelligent traffic accident liability assessment model is provided on the basis of the existing deep learning model, which is a very important and meaningful matter.
The invention relates to a block chain and road traffic patent in China, and discloses a block chain-based urban real-time traffic system and a block chain-based urban real-time traffic method (201710212277.4). The invention adopts a block chain as a core technology, utilizes an accounting book of the block chain to store and record the driving record of each automobile in a distributed manner, in the accounting book, each automobile is recorded in an automobile identifier mode, identifiers generate random character strings by adopting an encryption algorithm and change at regular time. The invention discloses a vehicle illegal behavior detection system based on a block chain and deep learning (202010308506.4). the method combines the block chain and the deep learning to realize the detection of the vehicle illegal behavior, but the system does not support the online upgrade of a model, so that the generalization of the model is restricted to a certain extent. In summary, in the aspect of algorithm, combining with the block chain technique, it is a big problem to be solved urgently to support the intelligent upgrade and update of the model.
Disclosure of Invention
In order to solve the problems, the invention provides a traffic accident responsibility judgment model based on a block chain and IVggNet by simultaneously adopting an advanced block chain technology on the basis of an SIFT algorithm and a VggNet model. In the aspect of model training data, data generated in the event of an accident are collected from multiple aspects such as vehicle-mounted GPS data, electronic monitoring data and vehicle networking data, so that the data are provided for network model training more comprehensively, and meanwhile, a block chain technology is adopted to classify, integrate, store and share massive and complex data. In a model algorithm processing module, SIFT is adopted to extract and enhance local features of an image, meanwhile, on the basis of an original VggNet model, a new loss function is provided to improve the discrimination accuracy and the generalization of the model, for accidents with misjudgment, the online upgrade of IVggNet is supported, and then the updated IVggNet is shared to a block chain, so that the model can be continuously updated and upgraded. To achieve the purpose, the invention provides a traffic accident responsibility judgment model based on a block chain and IVggNet, which comprises the following specific steps:
step 1, uploading data of the traffic accident to a block chain node, wherein the data comprises: vehicle-mounted GPS data, picture and video data shot by electronic monitoring equipment, Internet of vehicles data and the like;
further, the data of the traffic accident in the step 1 specifically include:
the vehicle-mounted GPS data comprises: the change data of the automobile speed, the change data of the braking acceleration and the like when an accident occurs; the electronic monitoring data includes: road condition pictures shot by a camera, videos of accidents, traffic flow space-time distribution maps, driving postures of drivers and the like; the car networking data comprises: control data, environmental data, operating condition data and the like.
Step 2, classifying and integrating the uploaded accident data by using a block chain technology and correspondingly storing the uploaded accident data;
furthermore, the uploaded accident data is classified, integrated and correspondingly stored, and the method is characterized in that:
when the uploaded data is shared by the public, the current node can be accessed to the public chain; on the contrary, when the data is classified as a private owner, the construction of the private chain can be selected, and the data is encrypted.
Step 3, taking data from the block chain by using the proposed IVggNet model, further mining the data by using the improved VggNet after SIFT feature extraction and enhancement, and finally making judgment on the accident;
further, the specific steps of processing and analyzing the data by the IVggNet model in the step 3 are as follows:
step 3.1, performing local feature extraction on the data acquired in the step 2 by using Scale-invariant feature transform (SIFT) to enhance the features of the picture, wherein the specific steps are as follows:
step 3.1.1, constructing a multi-scale space, wherein the method adopts a Gaussian convolution kernel to complete corresponding scale transformation, and the obtained scale space image L (x, y, sigma) expression is as follows:
wherein G (x, y, sigma) is a two-dimensional Gaussian function,representing a convolution operation, I (x, y) represents the input image, and G (x, y, σ) is specifically expressed as:
in the formula, x and y represent coordinates of an image, and σ is a scale factor.
And 3.1.2, detecting an extreme point in a scale space, accurately positioning and screening, wherein a Taylor function is selected to expand near the extreme point, a point with low contrast is removed, and then an unstable edge response point is removed by using a Hessian matrix.
Step 3.1.3, determining the main direction of the characteristic point, wherein the solving formula of the gradient g (x, y) and the direction theta (x, y) is as follows:
in the formula, L (x, y) is a gaussian image in which the feature point is located.
And 3.1.4, rotating the coordinate axis to the main direction of the characteristic points, and calculating the characteristic descriptors of the characteristic points.
Step 3.2, inputting the data obtained in the step 3.1 into an improved VggNet to further mine the data; the method comprises the following specific steps:
step 3.2.1, building an IVggNet model by utilizing a TensorFlow framework, wherein the model network structure is as follows: input layer-convolution layer 1-convolution layer 2-pooling layer 1-convolution layer 3-convolution layer 4-pooling layer 2-convolution layer 5-convolution layer 6-convolution layer 7-pooling layer 3-convolution layer 8-convolution layer 9-convolution layer 10-pooling layer 4-convolution layer 11-convolution layer 12-convolution layer 13-pooling layer 5-full connecting layer 1-full connecting layer 2-Softmax layer.
Step 3.2.2, training the IVggNet model by utilizing the existing training data set, and adding a regular penalty term, namely a specific loss function L, into the original loss functionIVggNetCan be expressed as:
in the formula, LcAnd LrRespectively representing a cross entropy loss term and a regular term, N is the number of samples, s(i)In order to be the actual sample label,for the label identified by the Softmax layer, λ is a regularization coefficient, and a convolution kernel weight coefficient is represented as wiThe number is m.
Step 3.2.3, reversely updating the weight coefficient of the IVggNet model by utilizing an Adam algorithm until the loss function LIVggNetThe convergence threshold 1e-5 is reached, at which point the model training is deemed complete.
Step 4, comparing the discrimination result in the step 3 with the discriminant result of the traffic police department, if the discrimination result is the same as the discriminant result of the traffic police department, skipping to the step 6, otherwise skipping to the step 5;
step 5, uploading the data of the accident and the judgment result of the traffic police department to a block chain model error correction training set, updating the IVggNet model when the sample data in the error correction training set reaches a set threshold value, and finally uploading the updated IVggNet model to a block chain to realize the sharing of the data and the model;
and 6, processing the traffic accident according to the result of the judgment.
The traffic accident responsibility judgment model based on the block chain and the IVggNet has the advantages that: the invention has the technical effects that:
1. the method collects data generated in the event of an accident from multiple aspects such as vehicle-mounted GPS data, electronic monitoring data, vehicle networking data and the like so as to provide data for network model training more comprehensively, and meanwhile, a block chain technology is adopted to classify, integrate, store and share massive and complex data;
2. according to the method, SIFT is adopted to extract and enhance local features of the image, and meanwhile, on the basis of an original VggNet model, a new loss function is provided to improve the model discrimination accuracy and the generalization;
3. the method supports online upgrade of the IVggNet for accidents with misjudgment, and then shares the updated IVggNet to the block chain, so that the model can be continuously updated and upgraded to improve the accuracy and the generalization of the model.
Drawings
FIG. 1 is a diagram of a model architecture of the present invention;
Detailed Description
The invention is described in further detail below with reference to the following detailed description and accompanying drawings:
the invention provides a traffic accident responsibility judgment model based on a block chain and IVggNe, and aims to realize intelligent responsibility judgment on traffic accidents.
FIG. 1 is a diagram of a model architecture of the present invention. The steps of the present invention will be described in detail below with reference to the structural diagrams.
Step 1, uploading data of the traffic accident to a block chain node, wherein the data comprises: vehicle-mounted GPS data, picture and video data shot by electronic monitoring equipment, Internet of vehicles data and the like;
further, the data of the traffic accident in the step 1 specifically include:
the vehicle-mounted GPS data comprises: the change data of the automobile speed, the change data of the braking acceleration and the like when an accident occurs; the electronic monitoring data includes: road condition pictures shot by a camera, videos of accidents, traffic flow space-time distribution maps, driving postures of drivers and the like; the car networking data comprises: control data, environmental data, operating condition data and the like.
Step 2, classifying and integrating the uploaded accident data by using a block chain technology and correspondingly storing the uploaded accident data;
furthermore, the uploaded accident data is classified, integrated and correspondingly stored, and the method is characterized in that:
when the uploaded data is shared by the public, the current node can be accessed to the public chain; on the contrary, when the data is classified as a private owner, the construction of the private chain can be selected, and the data is encrypted.
Step 3, taking data from the block chain by using the proposed IVggNet model, further mining the data by using the improved VggNet after SIFT feature extraction and enhancement, and finally making judgment on the accident;
further, the specific steps of processing and analyzing the data by the IVggNet model in the step 3 are as follows:
step 3.1, performing local feature extraction on the data acquired in the step 2 by using Scale-invariant feature transform (SIFT) to enhance the features of the picture, wherein the specific steps are as follows:
step 3.1.1, constructing a multi-scale space, wherein the method adopts a Gaussian convolution kernel to complete corresponding scale transformation, and the obtained scale space image L (x, y, sigma) expression is as follows:
wherein G (x, y, sigma) is a two-dimensional Gaussian function,representing a convolution operation, I (x, y) represents the input image, and G (x, y, σ) is specifically expressed as:
in the formula, x and y represent coordinates of an image, and σ is a scale factor.
And 3.1.2, detecting an extreme point in a scale space, accurately positioning and screening, wherein a Taylor function is selected to expand near the extreme point, a point with low contrast is removed, and then an unstable edge response point is removed by using a Hessian matrix.
Step 3.1.3, determining the main direction of the characteristic point, wherein the solving formula of the gradient g (x, y) and the direction theta (x, y) is as follows:
in the formula, L (x, y) is a gaussian image in which the feature point is located.
And 3.1.4, rotating the coordinate axis to the main direction of the characteristic points, and calculating the characteristic descriptors of the characteristic points.
Step 3.2, inputting the data obtained in the step 3.1 into an improved VggNet to further mine the data; the method comprises the following specific steps:
step 3.2.1, building an IVggNet model by utilizing a TensorFlow framework, wherein the model network structure is as follows: input layer-convolution layer 1-convolution layer 2-pooling layer 1-convolution layer 3-convolution layer 4-pooling layer 2-convolution layer 5-convolution layer 6-convolution layer 7-pooling layer 3-convolution layer 8-convolution layer 9-convolution layer 10-pooling layer 4-convolution layer 11-convolution layer 12-convolution layer 13-pooling layer 5-full connecting layer 1-full connecting layer 2-Softmax layer.
Step 3.2.2, training the IVggNet model by utilizing the existing training data set, and adding a regular penalty term, namely a specific loss function L, into the original loss functionIVggNetCan be expressed as:
in the formula, LcAnd LrRespectively representing a cross entropy loss term and a regular term, N is the number of samples, s(i)In order to be the actual sample label,for the label identified by the Softmax layer, λ is a regularization coefficient, and a convolution kernel weight coefficient is represented as wiThe number is m.
Step 3.2.3, reversely updating the weight coefficient of the IVggNet model by utilizing an Adam algorithm until the loss function LIVggNetThe convergence threshold 1e-5 is reached, at which point the model training is deemed complete.
Step 4, comparing the discrimination result in the step 3 with the discriminant result of the traffic police department, if the discrimination result is the same as the discriminant result of the traffic police department, skipping to the step 6, otherwise skipping to the step 5;
step 5, uploading the data of the accident and the judgment result of the traffic police department to a block chain model error correction training set, updating the IVggNet model when the sample data in the error correction training set reaches a set threshold value, and finally uploading the updated IVggNet model to a block chain to realize the sharing of the data and the model;
and 6, processing the traffic accident according to the result of the judgment.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, but any modifications or equivalent variations made according to the technical spirit of the present invention are within the scope of the present invention as claimed.
Claims (4)
1. A traffic accident responsibility judgment model based on a block chain and IVggNet comprises the following specific steps, and is characterized in that:
step 1, uploading data of the traffic accident to a block chain node, wherein the data comprises: vehicle-mounted GPS data, picture and video data shot by electronic monitoring equipment, Internet of vehicles data and the like;
step 2, classifying and integrating the uploaded accident data by using a block chain technology and correspondingly storing the uploaded accident data;
step 3, taking data from the block chain by using the proposed IVggNet model, further mining the data by using the improved VggNet after SIFT feature extraction and enhancement, and finally making judgment on the accident;
step 4, comparing the discrimination result in the step 3 with the discriminant result of the traffic police department, if the discrimination result is the same as the discriminant result of the traffic police department, skipping to the step 6, otherwise skipping to the step 5;
step 5, uploading the data of the accident and the judgment result of the traffic police department to a block chain model error correction training set, updating the IVggNet model when the sample data in the error correction training set reaches a set threshold value, and finally uploading the updated IVggNet model to a block chain to realize the sharing of the data and the model;
and 6, processing the traffic accident according to the result of the judgment.
2. The traffic accident accountability model based on the blockchain and IVggNet according to claim 1, wherein: the data of the traffic accident in the step 1 specifically include:
the vehicle-mounted GPS data comprises: the change data of the automobile speed, the change data of the braking acceleration and the like when an accident occurs; the electronic monitoring data includes: road condition pictures shot by a camera, videos of accidents, traffic flow space-time distribution maps, driving postures of drivers and the like; the car networking data comprises: control data, environmental data, operating condition data and the like.
3. The traffic accident accountability model based on the blockchain and IVggNet according to claim 1, wherein: and 2, classifying and integrating the uploaded accident data and correspondingly storing the uploaded accident data, wherein the method is characterized by comprising the following steps of:
when the uploaded data is shared by the public, the current node can be accessed to the public chain; on the contrary, when the data is classified as a private owner, the construction of the private chain can be selected, and the data is encrypted.
4. The traffic accident accountability model based on the blockchain and IVggNet according to claim 1, wherein: the specific steps of IVggNet model data processing and analysis in the step 3 are as follows:
step 3.1, local feature extraction is carried out on the data obtained in the step 2 by utilizing scale invariant feature transformation to enhance the features of the picture, and the specific steps are as follows:
step 3.1.1, constructing a multi-scale space, wherein the method adopts a Gaussian convolution kernel to complete corresponding scale transformation, and the obtained scale space image L (x, y, sigma) expression is as follows:
wherein G (x, y, sigma) is a two-dimensional Gaussian function,representing a convolution operation, I (x, y) represents the input image, and G (x, y, σ) is specifically expressed as:
in the formula, x and y represent coordinates of an image, and sigma is a scale factor;
step 3.1.2, detecting an extreme point in a scale space, accurately positioning and screening, wherein a Taylor function is selected to be expanded near the extreme point, a point with low contrast is removed, and then an unstable edge response point is removed by using a Hessian matrix;
step 3.1.3, determining the main direction of the characteristic point, wherein the solving formula of the gradient g (x, y) and the direction theta (x, y) is as follows:
in the formula, L (x, y) is a Gaussian image where the characteristic point is located;
step 3.1.4, rotating the coordinate axis to the main direction of the characteristic points, and calculating characteristic descriptors of the characteristic points;
step 3.2, inputting the data obtained in the step 3.1 into an improved VggNet to further mine the data; the method comprises the following specific steps:
step 3.2.1, building an IVggNet model by utilizing a TensorFlow framework, wherein the model network structure is as follows: input layer-convolution layer 1-convolution layer 2-pooling layer 1-convolution layer 3-convolution layer 4-pooling layer 2-convolution layer 5-convolution layer 6-convolution layer 7-pooling layer 3-convolution layer 8-convolution layer 9-convolution layer 10-pooling layer 4-convolution layer 11-convolution layer 12-convolution layer 13-pooling layer 5-full-connection layer 1-full-connection layer 2-Softmax layer;
step 3.2.2, training the IVggNet model by utilizing the existing training data set, and adding a regular penalty term, namely a specific loss function L, into the original loss functionIVggNetCan be expressed as:
in the formula, LcAnd LrRespectively representing a cross entropy loss term and a regular term, N is the number of samples, s(i)In order to be the actual sample label,for the label identified by the Softmax layer, λ is a regularization coefficient, and a convolution kernel weight coefficient is represented as wiThe number is m;
step 3.2.3, reversely updating the weight coefficient of the IVggNet model by utilizing an Adam algorithm until the loss function LIVggNetThe convergence threshold 1e-5 is reached, at which point the model training is deemed complete.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110046331.9A CN112749661A (en) | 2021-01-14 | 2021-01-14 | Traffic accident responsibility judging model based on block chain and IVggNet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110046331.9A CN112749661A (en) | 2021-01-14 | 2021-01-14 | Traffic accident responsibility judging model based on block chain and IVggNet |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112749661A true CN112749661A (en) | 2021-05-04 |
Family
ID=75651758
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110046331.9A Pending CN112749661A (en) | 2021-01-14 | 2021-01-14 | Traffic accident responsibility judging model based on block chain and IVggNet |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112749661A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118097493A (en) * | 2024-01-23 | 2024-05-28 | 常州金梧桐智能科技有限公司 | Intelligent processing method for monitoring video based on Internet of things |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017211136A1 (en) * | 2016-06-07 | 2017-12-14 | 中兴通讯股份有限公司 | Smart traffic management method and system, and computer storage medium |
CN109543588A (en) * | 2018-11-16 | 2019-03-29 | 广州慧睿思通信息科技有限公司 | Method, apparatus, system, service platform and the medium that traffic accident responsibility determines |
CN110070642A (en) * | 2019-03-22 | 2019-07-30 | 天津大学 | A kind of traffic accident responsibility appraisal procedure and device based on deep learning |
CN110689460A (en) * | 2019-10-09 | 2020-01-14 | 腾讯科技(深圳)有限公司 | Traffic accident data processing method, device, equipment and medium based on block chain |
CN111046212A (en) * | 2019-12-04 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Traffic accident processing method and device and electronic equipment |
CN111696015A (en) * | 2019-03-13 | 2020-09-22 | 阿里巴巴集团控股有限公司 | Accident processing method, device and system, computing equipment and storage medium |
-
2021
- 2021-01-14 CN CN202110046331.9A patent/CN112749661A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017211136A1 (en) * | 2016-06-07 | 2017-12-14 | 中兴通讯股份有限公司 | Smart traffic management method and system, and computer storage medium |
CN109543588A (en) * | 2018-11-16 | 2019-03-29 | 广州慧睿思通信息科技有限公司 | Method, apparatus, system, service platform and the medium that traffic accident responsibility determines |
CN111696015A (en) * | 2019-03-13 | 2020-09-22 | 阿里巴巴集团控股有限公司 | Accident processing method, device and system, computing equipment and storage medium |
CN110070642A (en) * | 2019-03-22 | 2019-07-30 | 天津大学 | A kind of traffic accident responsibility appraisal procedure and device based on deep learning |
CN110689460A (en) * | 2019-10-09 | 2020-01-14 | 腾讯科技(深圳)有限公司 | Traffic accident data processing method, device, equipment and medium based on block chain |
CN111046212A (en) * | 2019-12-04 | 2020-04-21 | 支付宝(杭州)信息技术有限公司 | Traffic accident processing method and device and electronic equipment |
Non-Patent Citations (3)
Title |
---|
TONG ZHANG 等: "A deep neural network-driven feature learning method for multi-view facial expression recognition", 《IEEE TRANSACTIONS ON MULTIMEDIA》, vol. 18, no. 12, pages 2528 - 2536, XP011634440, DOI: 10.1109/TMM.2016.2598092 * |
刘远航: "基于深度学习肺部CT图像分割算法研究", 《中国优秀硕士学位论文全文数据库医药卫生科技辑》, no. 2020, pages 060 - 145 * |
赵海涛 等: "基于深度学习的车联边缘网络交通事故风险预测算法研究", 《电子与信息学报》, vol. 42, no. 1, pages 50 - 57 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN118097493A (en) * | 2024-01-23 | 2024-05-28 | 常州金梧桐智能科技有限公司 | Intelligent processing method for monitoring video based on Internet of things |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20210197851A1 (en) | Method for building virtual scenario library for autonomous vehicle | |
Wang et al. | A Vision‐Based Video Crash Detection Framework for Mixed Traffic Flow Environment Considering Low‐Visibility Condition | |
CN109816024B (en) | Real-time vehicle logo detection method based on multi-scale feature fusion and DCNN | |
CN111368687B (en) | Sidewalk vehicle illegal parking detection method based on target detection and semantic segmentation | |
CN110304068B (en) | Method, device, equipment and storage medium for collecting automobile driving environment information | |
CN109993138A (en) | A kind of car plate detection and recognition methods and device | |
AU2021235756A1 (en) | Systems and methods for image-based location determination and parking monitoring | |
CN111428558A (en) | Vehicle detection method based on improved YO L Ov3 method | |
CN110991523A (en) | Interpretability evaluation method for unmanned vehicle detection algorithm performance | |
CN110516691A (en) | A kind of Vehicular exhaust detection method and device | |
US20230061710A1 (en) | Method and system for muck truck management in smart city based on internet of things | |
CN113591717A (en) | Non-motor vehicle helmet wearing detection method based on improved YOLOv3 algorithm | |
CN116205024A (en) | Self-adaptive automatic driving dynamic scene general generation method for high-low dimension evaluation scene | |
CN112785610B (en) | Lane line semantic segmentation method integrating low-level features | |
CN114550023A (en) | Traffic target static information extraction device | |
CN112749661A (en) | Traffic accident responsibility judging model based on block chain and IVggNet | |
CN114048536A (en) | Road structure prediction and target detection method based on multitask neural network | |
CN110765900A (en) | DSSD-based automatic illegal building detection method and system | |
CN116630702A (en) | Pavement adhesion coefficient prediction method based on semantic segmentation network | |
CN111832463A (en) | Deep learning-based traffic sign detection method | |
CN109190451B (en) | Remote sensing image vehicle detection method based on LFP characteristics | |
CN116386018A (en) | License plate character classification method and system | |
Wu et al. | Research on asphalt pavement disease detection based on improved YOLOv5s | |
Shen et al. | A lightweight method for small scale traffic sign detection based on YOLOv4-Tiny | |
CN110555425A (en) | Video stream real-time pedestrian detection method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210504 |
|
RJ01 | Rejection of invention patent application after publication |